DirectSound: Basics
By: Jack Hoxley
Written: May 2000
Download: DS_Simple.Zip (12kb)

Luckily for you, DirectSound at the basic level is extremely easy; so easy you could be mistaken for thinking something was wrong. If it makes you feel better it does get harder...

If you are familiar with DirectDraw you will know about surfaces - and how they hold graphical information. There is a similiar structure used in DirectSound, but they are called buffers. A buffer holds the sound data in memory so that you can access it, change it, play it and do various other things.

All sounds are in .Wav format, which tends to be the standard sound format for most things, Directsound can only use .Wav format sounds. If you want to create your own sound format, or want to use a compression method you will have to write some clever conversion code.

The DirectSound program revolves around the main DirectX object, which is used to create a Directsound object:

You may notice that the DirectSoundCreate has a blank parameter (""); this is only done for simplicity and should not be used in a proper program. This parameter allows you to specify which sound driver that you want to use for your application, specifying ("") allows DirectSound to choose the default driver. This is usually fine, but it is best to allow the user to specify which sound driver he/she wants to use; it is quite possible that an end user may have 5 or 6 drivers, and/or two sound cards (although unlikely) - How do you know that the default driver is A: The Best one B: Problem free C: Supports the features that you want. By allowing the user to select which one they want you can save yourself a lot of hassle when distributing the program.

Once the DirectSound engine has been initialised successfully you can load sounds into the buffers. First, though, you will need some sounds to play. This is the only part where DirectSound becomes complicated (so far) as good sounds can make a program, and bad sounds can destroy it. Proffesional game developers will have proper recording studios and expensive equipment for recording their sounds - and you can tell; most people reading this probably wont have a recording studio available to them. If you have a decent soundcard and a reasonable microphone then you can probably get away without the recording studio, but you'll have to be careful.
The other factor to the sound files is their size and quality. Wav files can be extremely large - 100's of megabytes for 5 minutes or good quality audio. Even if you are using a large-capacity medium, you probably won't want it to have this size of sound file on it; on top of that, it will probably take a while to load a 100Mb sound file into memory (there are ways around this - see Streaming from a file). You now have several choices:

1. Use Compression. This can be quite complicated, and the sound will need as much memory as it did when uncompressed.
2. You can Stream from a file - this involves running it from the hard drive, and not from memory
3. You can lower the quality. This is the best, simplest and safest option

Quality. There are 2 things that you can bare in mind when selecting the quality:

1. Bit depth - 8 or 16, 16 is better quality but larger.
2. Number of channels. 1 or 2 - mono or stereo. If you use mono you will not get any seperation between the right and left speakers, using stereo will be better quality, but again, the file siaze will increase.

Try and fit your sound files into either of these catergories for a simple decision:

1. If it is a beep/click/squeek or other sort of small noise you can get away with using the lowest quality; the average person will not know/hear the difference between an 8bit beep and a 16bit beep. If the sound fits here, make it 1 channel, 8 bit
2. If it is speech, or a sound where quality is important - usually where you need to pick out individual noises; make it high quality. 16 bit and 2 channels.

Then there is the second part to think about, you need to specify these setting when CREATING the sound in your sound program. It won't make any difference if the sound quality is saved as 8bit/1channel and Directsound tries to play it at 16bit/2channel, nor will it make any difference to the file size.

This is a similiar structure to that used by DirectDraw when creating a surface - should you have knowledge of that.

Finally, we can play and stop our sound. This part requires no explanation - it's so easy:

Finished!

You can download a complete working example fromt he top of the page, or you can get it from the downloads page